<rellla>
looks like aw wants to "open" themselves. at least there is a login prompt at the download area...
<mnemoc>
the first is for a big logo, with small letters. and the second for an small logo (favicon) with proportionally larger letters
<oliv3r>
i'll gimp it
<mnemoc>
svg -> inkspace ;-)
<oliv3r>
oh yeah, hmm, i think i have that installed at home
<oliv3r>
but 0 knowledge on how to use it :p
<mnemoc>
oliv3r: that's another nice to know web service. filetea.me acts as secure https-based file transfer proxy. html5-based, nothing is cached/stored and no accounts are needed
<oliv3r>
rellla: doubt that. From some of the commits in the wiki, the page has been up for a few months, and it's mostly all in chinese
<oliv3r>
mnemoc: probably a lot faster then privproxy +tor
<mnemoc>
this is for people to share with people. not to proxy a url
<rm>
mnemoc, still have to keep the browser tab open?
<oliv3r>
mnemoc: how are we gonna handle the register lists. they seem to be 99% identical (for A10 and A13 anyway). have a generic list and add notes which chip has what? or duplicate them (god no)
<mnemoc>
rm: yes, you close it, it's gone
<oliv3r>
mnemoc: when I have a few hours to spare, i'll ask you again for the svg's
<mnemoc>
oliv3r: I would actually move A10/IP to just IP
<mnemoc>
oliv3r: ok
<oliv3r>
mnemoc: so 'merge' possible chips together where possible
<mnemoc>
oliv3r: yes
<oliv3r>
(in terms of register lists)
<oliv3r>
move the pages at that time
<oliv3r>
?
<mnemoc>
yes
<mnemoc>
or all now. doesn't really matter
<oliv3r>
i think you have to move the pages?
<oliv3r>
i can fix the links I suppose today
<mnemoc>
adding a note "this is based on A10 specs" or so
<mnemoc>
oliv3r: I think you can move them too...
<mnemoc>
1m
<oliv3r>
and i'll ad a collomn 'for devices'
<mnemoc>
oliv3r: just check and you are already admin. so you can move, delete, ban, etc etc
<oliv3r>
power!
<oliv3r>
I'll re-login to see if that changes anything. I don't see any new options/buttons yet ;)
<mnemoc>
there is an arrow between [view history] and the search box
<rellla>
oliv3r: in fact i doubt that, too :( but last change is from today, and there have been at least a few contributors...
<oliv3r>
ah yes, so very obvious and small :) i did relog in, so that may have made a difference
<oliv3r>
rellla: what is the oldest commit
<mnemoc>
oliv3r: it hasn't. you've been admin at least for 90 days as the log doesn't show the change anymore
<oliv3r>
lol oh
<rellla>
oliv3r: first main page version is 2012/03/08
<oliv3r>
so since march
<oliv3r>
a year old
<oliv3r>
i found that the SD Boot page was from end of november 2012 and an exact copy from the rhombus-tech site :)
paulk-desktop has quit [Ping timeout: 248 seconds]
paulk-desktop has joined #linux-sunxi
anunnaki has joined #linux-sunxi
<hramrach>
yes, only works while the tab is open
<hramrach>
hmm, somehow filetea fails for me
<hramrach>
used to work
<oliv3r>
failtea! :p
<hramrach>
sucks you still need a server-in-the-middle to send a file
<oliv3r>
you can send files via IM or IRC (dcc) but with nat and god knows what, that's more trouble then it's worth
<hramrach>
somehow internet speeds gorw but the simple operation of transferrign a file from one PC to another .. for an ordinary Joe User it's still easiest to put it on a flash drive and drive it to the other PC in a car ;-)
<hramrach>
note that IM and IRC do use server-in-the-middle
<hramrach>
the final dcc may not but it's server-moderated protocol still
<oliv3r>
give us global IPv6 and the problem can be solved
<oliv3r>
but then you turn every client into a server
<hramrach>
that's the point
<hramrach>
central servers should die
<hramrach>
they are useful in some cases but in many other they get in the way
<hramrach>
anyone understands ARM assembly?
<oliv3r>
i wonder how we use the internet (webpages) without serveres :)
<hramrach>
there are ways
<oliv3r>
p2p the internet!
<oliv3r>
gone are servers
<hramrach>
but the question is not how we do everything without servers but why do you need damn hosting for web pages when your home pc is powerful enough to serve them
<oliv3r>
always on?
<oliv3r>
i server plenty of webpages from my home connection
<oliv3r>
but i have a server at home :)
<oliv3r>
it's all to do with 99.9% uptime
<oliv3r>
there's people who server stuff from home, and add a note 'only available from 9:00 - 23:00, because i sleep otherwise
<hramrach>
;-)
<oliv3r>
what you'd want, i sall those modems/routers, that run linux as it is most of the time anwywa, to have a basic webserver so you can use it to serve files or share files with :)
<hramrach>
with an ARM box or a router you can have an always-on machine
<hramrach>
but the thing is unless you are a geek you would not know how to serve web off that
egbert_ is now known as egbert
<hramrach>
and it's still quite useless for the conceptually very simple 'send this file to that other pc"
<oliv3r>
well manufactures could pre-install something like samba on it with a /public_html folder ready available, just drop files on it and done, that's server from your IP
<oliv3r>
not that hard
<oliv3r>
but usaully there's not enough storage in said unit
<oliv3r>
and ... ISP don't really want you to od that aynway
<hramrach>
there is still that intermediate step
<hramrach>
you upload the file to a server (your server in this case), and other person downloads from there
<oliv3r>
we where talking about 'homepage'
<hramrach>
home pages are doable with what we have, yes
<hramrach>
except you need DNS hosting then
<oliv3r>
decentralized DNS service
ganbold_ has joined #linux-sunxi
<hramrach>
hmmm, I don't understand but I guess the patch does the right thing
<hramrach>
the math in ARM assembly is really awkward
<libv>
what do you need arm assembly for?
<mnemoc>
neon optimized algorithms?
<libv>
let libc deal with that.
<libv>
i still have to see "optimized hand-written assembly" for any project be anything more than someones short term petpieve, one that tends to get in the way 6 months on
<mnemoc>
but it exposes the usb device port externally :)
<arete74>
i want an cubieboard A20!
<ganbold_>
mnemoc: expensive is this a20 devices
<mnemoc>
dhl shipping included
<hramrach>
libv: to understand what the object file contains. not so much interested in hand-crafting new object files when a compiler does decent job automagically
<mnemoc>
but shipping is from .cn so it implies dhl-punishment on delivery
<hramrach>
still there are code pieces like media encoders/decoders that benchmark better when hand-coded in assembly
<hramrach>
that mini pc looks nice but £94.99 ..
<libv>
hramrach: what object file? neon optimizations, to what?
<libv>
hramrach: my experience on unichrome: playing cpu specific cache games made upload to FB faster
<libv>
loads of tinkering, and every different cpu version is different.
<hramrach>
well, neon optimizations are rather theoretical use of hand-crafted assembly but eg. libjpeg-turbo has patches for that
<libv>
or you could also just set up the northbridge memory controller correctly and get the same bandwidth as the cpu has
<hramrach>
indeed, different CPUs handle NEON differently so the benchmarks will probably have different results on A8 and A9
<hramrach>
but what I was looking at was a patched libmali
<hramrach>
and I can't tell how the hell does the patch work even looking at the disassembled object file
<libv>
what patched libmali?
<libv>
the patch i did to stop it from falling over?
<hramrach>
one that calls XInitThreads() in _init()
<libv>
my change is the best we can do
<libv>
XCloseDisplay makes it fall over.
<libv>
attempting to tear down X resources after XCloseDisplay is simply not smart
<libv>
and the libegl destructor would of course only happen on exit()
<hramrach>
oh, you have another patch
<hramrach>
but the one I have makes gnome-shell not fail on Debian
<hramrach>
and other stuff reportedly
<libv>
ok
<hramrach>
do you have bsdiff for the atexit patch?
<libv>
it's in git
<hramrach>
I guess I can diff it from there, thanks
<oliv3r>
is that an A20 based device?!?!?!
<hramrach>
yes, it is
<libv>
i basically ret in the destructor, skipping all the pointless nonsense that is attempted afterwards
<oliv3r>
mnemoc: stop distracting me :p
<oliv3r>
while I admire their use of composite out, svideo out is so much better (in quality and its backwards compatble!)
<mnemoc>
:)
<oliv3r>
so cloudsto is mele from the UK
<oliv3r>
if a company in th eUK has it right now, where are the chinese companies selling it?
<ganbold_>
uk is expensive
<oliv3r>
I find it hard to imagine, that 'some' UK company is the first with A20 devices. Unless these are 'samples'
<oliv3r>
ah, es, that is true
<oliv3r>
highest margins
<ganbold_>
or whole europe
<libv>
oliv3r: they are undoubtedly chinese rebadged devices
<libv>
and they are not shipping yet.
drachensun has joined #linux-sunxi
<ganbold_>
just checked tmall and taobao nothing on a20 etc
<oliv3r>
as expected
<oliv3r>
this might just aswell be some prototype markup, so could be several weeks/months out
<hramrach>
that wireless keyboard they sell is kinda nice
<oliv3r>
how many bytes is a sector generally on an SD card, and at which sector does uboot-SPL start.
<hramrach>
sector is 512 bytes
<hramrach>
it's an artifical value
<oliv3r>
a sector on an HDD is, but on an SD card?
<oliv3r>
well 512 or 4k nowadays
<hramrach>
the actual block size is something like 2k or 4k (possibly larger with newer chips), erase block 16k+
<hramrach>
but the interface is in 512 byte sectors
<hramrach>
the kernel interface is in 1k? blocks but usable io size is 4k (page?)
<hramrach>
again, the 1k blocks are fabrication which some kernel interfaces use to measure device capacity totally unrelated to anything below or above
<hramrach>
other lower-level interfaces use the sector count
<hramrach>
oliv3r: as for u-boot look at the wiki. there is a page which has details about the layout or link to an u-boot page with details
wingrime has joined #linux-sunxi
ZaEarl has quit [Ping timeout: 272 seconds]
paulk-desktop has quit [Ping timeout: 248 seconds]
rellla has quit [Remote host closed the connection]
<ZaEarl>
Makefile says "./gen_check_code ./resume1.bin ./super/resume/resume1.code", but I can't find resume1.bin
ganbold_ has quit [Remote host closed the connection]
<mnemoc>
ZaEarl: allwinner SDK stuff assumes you use their ./build.sh
<ZaEarl>
I'll try that
<ZaEarl>
any idea what sun6i_fiber or sun6i_dragonboard are?
<mnemoc>
no idea
<mnemoc>
diff them
<mnemoc>
the defconfig I mean
<ZaEarl>
arm-linux-gnueabi-gcc: error: unrecognized command line option ‘--min_array_alignment=4’
<ZaEarl>
arm-linux-gnueabi-gcc: error: unrecognized command line option ‘--no_unaligned_access’
<mnemoc>
ZaEarl: so they want a different gcc version
<ZaEarl>
odd. I have the latest 4.7.2. gcc dropped those options?
<mnemoc>
ZaEarl: you are the first to try that source drop.
<ZaEarl>
not very successfully
<mnemoc>
expectable
<ZaEarl>
tru
<ZaEarl>
mnemoc, where did the source come from?
<mnemoc>
an annonymous fellow who wanted to comply with the GPL
<ZaEarl>
good for him
andy has quit [Ping timeout: 245 seconds]
<mnemoc>
got two urls with tarballs of GPL code saying A20 and A31.... I won't reject them :)
<ZaEarl>
might he know how to build it?
<mnemoc>
the A1x SDKs only worked with gcc 4.4
<ZaEarl>
I think I have 4.4 in my chroot. I'll give it a try
<wingrime>
mnemoc: uboot board config requset DRAM settings only ?
<wingrime>
I wanna add support for my board
<mnemoc>
wingrime: get the .bin converted to .fex, commit it to sunxi-boards. run a10meminfo in android as root. use the output to complete/verify the [dram_para] section of the .fex. commit it again
<mnemoc>
wingrime: and then use sunxi-tools's fexc -O u-boot to generate the board.c for u-boot
<wingrime>
mnemoc: Default or fixed fex? It not possible use default fex with modern kernels....
<mnemoc>
wingrime: first the default .fex, then the .fex with fixed [dram_para], then disable the foo_used = 1 which aren't really used
<mnemoc>
wingrime: three commits, and submit to linux-sunxi ;-)
<mnemoc>
wingrime: what board is it?
<wingrime>
mnemoc: Explay informer 801 (I have rebrended version)
<wingrime>
tablet
<mnemoc>
wingrime: you can also add a page for it in the wiki, including specs, a picture of the PCB and if possible the uart pads ;-)
<wingrime>
mnemoc: I have anouser tablet on a13
<mnemoc>
you can do the same with that one :)
<wingrime>
mnemoc: UART pads already found
<mnemoc>
wiki wiki :)
<wingrime>
mnemoc: first time I wanna make support, wiki on second order
<wingrime>
also near TX/RX there are 4 pins, can it be JTAG ?
<wingrime>
pins=test points
<wingrime>
mnemoc: dram_para in fex always invalid ?
paulk-desktop has joined #linux-sunxi
<mnemoc>
wingrime: they usually leave NULL fields for those things that livesuit can detect
<mnemoc>
but we don't have that power. also probing the dram on every boot isn't such a good idea
<wingrime>
mnemoc: In wiki there is one more way to see dram configs using u-boot and uart this way still usable ? I prefer use use uart when tablet disasembled.
<mnemoc>
wingrime: it's not reliable
<mnemoc>
wingrime: the u-boot methods is to dump boot0 from sram
<mnemoc>
but sometimes it's not there anymore, or not complete
<wingrime>
mnemoc: a10meminfo runable on debian ?
<mnemoc>
wingrime: a10meminfo reads the dramc registers. if you don't boot from nand you won't have it properly initialized
<mnemoc>
if you boot debian from nand using the stock boot0/boot1 stuff, sure, you can use a10meminfo from debian